jperry@UNIX.SRI.COM.UUCP (05/28/87)
It is interesting to note that you list Lisp, Bliss, and FP in the "clear, simple language" category that you imply I am supporting. Actually I find Lisp, Bliss, and FP to be abominations. Their simple syntax is a misleading disguise for they lead to programs whose semantics are virtually unintelligible. There is nothing clear to me at all about the intent of a Lisp, Bliss, or FP program upon brief inspection of the code. Oddly enough, programs written in declarative languages (like Hope, FP, or Prolog) are the most obfuscated of all --- the clauses of which they consist seem quite clear when described in English but when written in an actual syntax with anonymous variables, cut/fail operators, and recursive rules the reader might know what's supposed to happen conceptually but goes bonkers trying to figure out HOW THE COMPUTER GOES ABOUT solving the problem. AI languages in general should be suspect on strictly empirical grounds. The programs written in those languages are known for being far buggier than normal commercial software and far more unreliable when the programs are asked to operate beyond strictly tested limits. It's no secret that AI is fast approaching its own version of a "nuclear winter" --- even top-notch companies like Quintas (the Prolog people) are barely staying afloat. Yet I have not heard a single soul conjecture that just maybe the hoary tools used to generate these programs might have something to do with their colossal failures. Again, I claim, as I have in many past memos and will probably continue to do so, that it is in the nature of the programmer's ego to use the most arcane, "in-group" tools available to solve a problem. God forbid that it would be revealed that something is easier than it looks rather than harder. John Perry
PETERSC0@VUCTRVAX.BITNET.UUCP (05/28/87)
> From: arpa%"jperry@sri-unix.arpa" 27-MAY-1987 22:48 > It is interesting to note that you list Lisp, Bliss, and FP in the > "clear, simple language" category that you imply I am supporting. Actually > I find Lisp, Bliss, and FP to be abominations. Their simple syntax is > a misleading disguise for they lead to programs whose semantics are > virtually unintelligible. Hmmm... It seems to me (philosophically speaking), that almost any language can be either readable or nonreadable depending on the style of the person doing the coding and on the familiarity of the reader with the language. Pascal enforces a (very) few rules of style, as does C. I can write things in either language which are readable or unreadable depending on my intent. My lisp has had to be readable for classwork, but I have seen a lot of strange-looking code out there. Doesn't it all depend on the coder in the final analysis? -Chris